I have launched a project on bias in speaker verification with the Trustworthy AI working group of the Mozilla Foundation, and you are invited to join us.
If you're interested to get involved, you can find out more about the project here
We are a small, interdisciplinary group of collaborators working on the
topic, and welcome people from different backgrounds. We are hosting a
hackathon tomorrow (Saturday, 19 November 2021, 6pm - 9pm UTC+2).
What is speaker verification and why does it matter if it's biased? Speaker verification is a form of voice-based biometric identification. You can think of it as face recognition using the voice. Over the last year you may have noticed that speaker verification is popping up in voice interfaces around you - your bank trying to switch you over to voice-based identification, call centres (where you are probably unaware of it being used), and of course voice assistants on your phone. Currently no investigations have been made into bias in speaker verification. Knowing that face recognition technology has major challenges around discrimination, it is urgent and important that we scrutinise the technology.
free to reach out if you have any questions.
Happy Friday :)